Latin square design

I've got an experiment that I'm designing, and I've got the following issue. I'd like to present a video before a series of questions, but I'd like to choose from sets of videos to present. So, I've got four sets of four videos each, and I'd like to present one out of each set at random, and I'd like each of the set representatives to be presented in random order. Does that make sense?

Set 1 (a,b,c,d)
Set 2 (e,f,g,h)
Set 3 (i,j,k,l)
Set 4 (m,n,o,p)

The questionnaire should randomize sets 1-4, and when a set is presented, a single video is chosen at random from that set. So,

Set2 g
Set3 l
Set1 d
Set4 p

and

Set 3 k
Set 1 b
Set 4 o
Set 2 f

are both possible presentations.

Is this kind of randomization possible in MediaLab?

Yes, you can assign a constant BGR (between groups randomization) value to all the videos within a single set like this:

Set 1 (a=1,b=1,c=1,d=1)
Set 2 (e=2,f=2,g=2,h=2)
Set 3 (i=3,j=3,k=3,l=3)
Set 4 (m=4,n=4,o=4,p=4)

That will keep the sets together but will randomize the set order (e.g., set 2, set 3, set 1, set 4).

Then assign constant WGR ([I]within[/I] groups randomization) value to all the videos within each set in the same manner:

Set 1 (a=1,b=1,c=1,d=1)
Set 2 (e=2,f=2,g=2,h=2)
Set 3 (i=3,j=3,k=3,l=3)
Set 4 (m=4,n=4,o=4,p=4)

That will randomize the order of the videos within each set, e.g., Set 1 (c, d, a, b).

The trick now is to present just one video from each set and not all four. Pretty easy--just insert an instruction screen at the end of each block. For every item within the block--have it skip automatically to that end item. This way, no matter which stimulus is shown first, MediaLab will then skip to the end of the set--voilà!

Ok, it might be way easier to see an example.. Take a look at the illustration I've posted below.

-Blair

Thanks,

I ended up figuring out something pretty much like that, and it works, so I'm not gonna monkey with it anymore. I think I was making it harder than it needed to be.

Ok, I've got a new wrinkle in the design. The way I succeeded in randomizing was to create four questionnaires. Each questionnaire contains a set of videos. Using the WGR option, the videos within each questionnaire are randomized, and they all skip to the questions, so only one video is presented. In the main experiment the questionnaires are randomized, so that the questionnaires are presented in random order. This seemed like the best way to keep the questionnaires for each video set with that video set.

However, now, it looks like we want to have some of the instructions for the questionnaires to be presented only for the first questionnaire presented, whichever one that happens to be.

Since I don't know which questionnaire is going to be first, I've got to find a way to skip the instructions after one video has been presented.

I've thought about doing this using the responses.xls. If a video is presented, it gets an "ok" value, and if it isn't it seems to end up with a "99" value. I wanted to find a way to have the skipto values for the screens prior to the instructions to be given a skip value if any of the videos have an "ok" value. So, in responses.xls, there would be a row for each video filename (VideoA, VideoB,etc.) and for each of the pre-instruction screens, the variable would have an "if" function that would enter a skip value if VideoA=ok or VideoB=ok or, etc. In theory, that might work, but when the next person runs the experiment, responses.xls will still have the values from the previous run, right? So, there will be 4 videos that have the "ok" value, and everything will get skipped.

Is there a way to reset responses.xls without having to manually clear it between experiments?

Wow, clever solution! No worries, the responses.xls file does not save the values from the session except as a copy in the "reports" subfolder of the data folder. It will have only those values you have saved in there from the start.

Oh, ok. My mistake. For some reason, I thought that responses.xls would keep the changed values from session to session. That's cool. It looks like we may have decided to simply have a preliminary practice session before the video presentations, so we can have the instructions there, and there isn't any real need for the responses.xls solution.

I may try it out, anyway, just to see if it works like I think it should. Just for kicks, if nothing else.

Thanks!